On the Relationship Between Recognition Familiarity and Perceptual Fluency: Evidence for Distinct Mnemonic Processes
Total Page:16
File Type:pdf, Size:1020Kb
Acta Psychologica 98 (1998) 211±230 On the relationship between recognition familiarity and perceptual ¯uency: Evidence for distinct mnemonic processes Anthony D. Wagner *, John D.E. Gabrieli Department of Psychology, Stanford University, Jordan Hall, Bldg. 420, Stanford, CA 94305-2130, USA Abstract Fluent reprocessing of perceptual aspects of recently experienced stimuli is thought to sup- port repetition priming eects on implicit perceptual memory tests. Although behavioral and neuropsychological dissociations demonstrate that separable mnemonic processes and neural substrates mediate implicit and explicit test performance, dual-process theories of memory posit that explicit recognition memory judgments may be based on familiarity derived from the same perceptual ¯uency that yields perceptual priming. Here we consider the relationship between familiarity-based recognition memory and implicit perceptual memory. A select re- view of the literature demonstrates that the ¯uency supporting implicit perceptual memory is functionally and anatomically distinct from that supporting recognition memory. In con- trast to perceptual ¯uency, recognition familiarity is more sensitive to conceptual than to per- ceptual processing, and does not depend on modality-speci®c sensory cortices. Alternative possible relationships between familiarity in explicit memory and ¯uency in implicit memory are discussed. Ó 1998 Elsevier Science B.V. All rights reserved. PsycINFO classi®cation: 2343 Keywords: Explicit memory; Declarative memory; Implicit memory; Nondeclarative memory; Recognition memory; Priming; Process dissociation; Dual-process theories * Corresponding author. Present address: Department of Psychology, Harvard University, Room 854, William James Hall, Cambridge, MA 02138, USA. Tel.: +1 617-495-3856; fax: +1 617-496-3122; e-mail: [email protected]. 0001-6918/98/$19.00 Ó 1998 Elsevier Science B.V. All rights reserved. PII S 0 0 0 1 - 6 9 1 8 ( 9 7 ) 0 0 0 4 3 - 7 212 A.D. Wagner, J.D.E. Gabrieli / Acta Psychologica 98 (1998) 211±230 1. Introduction Experience is recorded as multiple mnemonic representations, with remembrance of the past corresponding to retrieval of these representations. Memory can be probed using a variety of retrieval cues, generally classi®ed as explicit (or direct) and implicit (or indirect), with dierent cues eliciting the recovery of dierent kinds of representations. Explicit tests refer directly to an episode and require conscious recollection of an aspect of the episode. Recognition, for example, requires a judg- ment of whether a test stimulus was encountered in a particular spatiotemporal learn- ing context. Implicit tests, in contrast, make no reference to any particular episode. Rather, memory is measured indirectly as a change in test-phase performance that is attributable to a particular study-phase experience. One kind of implicit measure is repetition priming, a facilitation or bias in task performance due to prior processing of a stimulus. Priming is thought to re¯ect an enhancement in the ¯uency with which a test-stimulus is processed. Priming can be perceptual when it re¯ects ¯uent repro- cessing arising from prior processing of stimulus form, or conceptual when it re¯ects ¯uent reprocessing arising from prior processing of stimulus meaning. Behavioral studies of healthy adults have demonstrated functional dissociations between performance on explicit and implicit tests and on perceptual and conceptual implicit tests (for reviews see, Richardson-Klavehn and Bjork, 1988; Roediger and McDermott, 1993). For example, recognition accuracy is enhanced by conceptual encoding and is often unaected by changes in perceptual form, whereas perceptual priming is greatest when study and test perceptual forms match and is unaected by manipulations of conceptual encoding (e.g., Jacoby, 1983; Jacoby and Dallas, 1981; Winnick and Daniel, 1970). Similarly, conceptual and perceptual implicit memory are dissociable using manipulations that vary the extent of conceptual encoding or the match between study and test perceptual form (e.g., Blaxton, 1989; Srinivas and Roediger, 1990). These dissociations suggest that these implicit and explicit mea- sures index functionally distinct processes and representations. Neuropsychological and neuroimaging studies have demonstrated anatomic disso- ciations between performance on explicit, implicit perceptual, and implicit conceptual tests. For example, damage to medial temporal and diencephalic structures impairs performance on explicit, but not implicit, memory tests (for reviews see, Schacter et al., 1993; Squire, 1992; Squire et al., 1993). In contrast, lesions of modality-speci®c sensory cortices selectively impair implicit perceptual memory (Fleischman et al., 1995; Fleischman et al., in press; Gabrieli et al., 1995; Keane et al., 1995; Vaidya et al., in press), whereas damage to temporal, parietal, and frontal cortices impairs im- plicit conceptual memory (e.g., Gabrieli et al., 1994; Keane et al., 1991; Salmon et al., 1988). Neuroimaging studies provide convergent evidence revealing: (a) activation in medial temporal structures during recognition and cued recall (e.g., Buckner et al., 1995; Gabrieli et al., 1997; Schacter et al., 1996; Squire et al., 1992), but not during perceptual priming (Schacter et al., 1996); (b) decreased activation in extrastriate cor- tex associated with visual priming (e.g., Buckner et al., 1995; Squire et al., 1992); and (c) decreased activation in left inferior frontal cortex associated with conceptual prim- ing (e.g., Demb et al., 1995; Raichle et al., 1994; Wagner et al., in press). Thus, ana- A.D. Wagner, J.D.E. Gabrieli / Acta Psychologica 98 (1998) 211±230 213 tomic and functional dissociations suggest that distinct processes and neural sub- strates mediate explicit, implicit perceptual, and implicit conceptual memory (e.g., Cohen and Squire, 1980; Gabrieli et al., 1994; Schacter, 1992; Squire, 1992). Although it is widely held that explicit and implicit tests index unique mnemonic processes, it is less clear whether performance on these measures may also rely on shared processes. One class of memory models ± dual-process models of recognition memory ± posit that a common process supports both recognition judgments and perceptual priming. From the dual-process perspective, recognition judgments can be based on two distinct processes, recollection and familiarity. Recollection is thought to consist of the conscious remembrance of some aspect of a prior experi- ence. Familiarity, in contrast, is thought to be a subjective sensation that occurs when ¯uent processing of a stimulus is unconsciously attributed to past experience (e.g., Atkinson and Juola, 1974; Gardiner, 1988; Jacoby, 1983, 1991; Jacoby and Dallas, 1981; Mandler, 1980, 1991). It has been proposed that recognition familiarity primarily derives from the perceptual ¯uency that supports implicit perceptual mem- ory. On implicit perceptual tests, ¯uent reprocessing of perceptual aspects of previ- ously experienced stimuli yields perceptual priming (e.g., Jacoby and Dallas, 1981). On explicit recognition tests, the same perceptual ¯uency is thought to produce a sense of familiarity that can be used heuristically to discriminate studied from un- studied words (e.g., Gardiner, 1988; Gardiner and Java, 1990; Gardiner and Parkin, 1990; Jacoby, 1983, 1991; Jacoby and Dallas, 1981; Mandler, 1980; Rajaram, 1993; Yonelinas et al., 1995). Support for the assertion that perceptual ¯uency mediates recognition familiarity comes from studies of recognition memory where ¯uency of test-item processing was systematically manipulated and the eects of these manipulations on recognition judgments was measured (e.g., Johnston et al., 1985, 1991; Kelley et al., 1989). Most studies in this vein have modulated ¯uency of test word processing by varying the density of visual noise masks (e.g., Whittlesea, 1993; Whittlesea et al., 1990) or by providing a brief masked priming presentation of a test word just prior to its occur- rence (e.g., Forster, 1985; Jacoby and Whitehouse, 1989). Reductions in the density of a noise mask and presentation of test-item primes serve to increase participants' willingness to judge a test item as previously encountered, regardless of whether or not it actually had been studied, or to judge a test item as previously encountered for a longer than for a shorter duration, regardless of whether the study presentation was long or short. To the extent that these manipulations wield their eects by in¯u- encing ¯uency of test-item perceptual processing, then these results suggest that per- ceptual ¯uency is used as an attributional source for recognition. These results, how- ever, do not inform us as to whether this process is the same as the long-term ¯uency process that supports perceptual priming. Other evidence indicates that familiarity-based recognition is modulated by ma- nipulations of test-item conceptual processing. For example, in a study by Whittlesea (1993), recognition test words were embedded at the end of conceptually related or unrelated sentence contexts. When the sentence context was conceptually predictive of the test word, participants were more likely to judge the word as having been pre- viously encountered, regardless of whether or not the word had been studied. 214 A.D. Wagner, J.D.E. Gabrieli / Acta Psychologica 98 (1998) 211±230 Furthermore, these experiments reveal that such conceptual processing manipula- tions can have considerably larger eects on